Synthesis and transport of glycoproteins and proteoglycans in the apical and basolateral secretory pathways of epithelial
MDCK cells
Thesis submitted for the degree of Philosophiae Doctor By
Tilahun Tolesa Hafte
DEPARTMENT OF MOLECULAR BIOSCIENCES FACULTY OF MATHEMATICS AND NATURAL SCIENCES
UNIVERSITY OF OSLO NORWAY
2011
© Tilahun Tolesa Hafte, 2011
Series of dissertations submitted to the
Faculty of Mathematics and Natural Sciences, University of Oslo No. 1083
ISSN 1501-7710
All rights reserved. No part of this publication may be
reproduced or transmitted, in any form or by any means, without permission.
Cover: Inger Sandved Anfinsen.
Printed in Norway: AIT Oslo AS.
Produced in co-operation with Unipub.
The thesis is produced by Unipub merely in connection with the
thesis defence. Kindly direct all inquiries regarding the thesis to the copyright holder or the unit which grants the doctorate.
2
ACKNOWLEDGEMENTS...3
ABBREVIATIONS...5
LIST OF PAPERS...7
SUMMARY...8
1. Introduction ... 11
1.1 EPITHELIAL CELLS ... 11
1.1.1 MDCK cells ... 12
1.2 THE SECRETORY PATHWAY ... 13
1.2.1 Transport from the ER to the Golgi apparatus ... 13
1.2.2 The Golgi apparatus ... 15
1.2.3 Protein modifications in the Golgi apparatus ... 15
1.2.4 Protein journey through the Golgi apparatus ... 16
1.2.5 Sorting of proteins in the secretory pathway of epithelial cells ... 19
1.3 SORTING SIGNALS FOR THE JOURNEY BEYOND THE GOLGI APPARATUS ... 20
1.3.1 Basolateral sorting signals ... 20
1.3.2 Apical sorting signals ... 22
2. Glycoproteins and proteoglycans ... 25
2.1 SYNTHESIS OF N- AND O-LINKED GLYCANS ... 25
2.2 SYNTHESIS OF PROTEOGLYCANS (PGS) AND GLYCOSAMINOGLYCANS (GAGS) ... 31
2.2.1 Chondroitin sulfate synthesis ... 34
2.2.2 Heparan sulfate synthesis ... 35
3
4. Aims of study ... 39
5. Summary of Papers ... 40
PAPER I ... 40
PAPER II ... 41
PAPER III ... 42
6. Discussion ... 44
6.1 GAG MODIFICATIONS OBTAINED IN THE APICAL AND BASOLATERAL SECRETORY PATHWAYS OF EPITHELIAL MDCK CELLS, AND THEIR EFFECT ON POLARIZED SORTING ... 45
6.2 ANY POSSIBLE PHYSIOLOGICAL ROLE OF DIFFERENTIAL SULFATION IN EPITHELIAL CELLS? ... 50
6.3 FURTHER IMPLICATIONS OF EARLY SEGREGATION OF THE APICAL AND BASOLATERAL SECRETORY PATHWAYS ... 51
6.4 ANALYSISIS OF N-GLYCANATED RGH WITH AND WITHOUT GAG BINDING DOMAIN ... 52
References ... 54
4
ACKNOWLEDGEMENTS
The work presented in this thesis has been carried out at the Department of Molecular Biosciences, Faculty of Mathematics and Natural Sciences, University of Oslo from 2006 to 2010. The financial support for this work was provided by the Faculty of Mathematics and Natural Sciences, University of Oslo.
First of all, I would like to thank my supervisor Professor Kristian Prydz for introducing me to this fascinating field of proteoglycans and for his skilful guidance and help throughout this study. His door has always been open to me whenever I was in need of his help. I really appreciate his ‘non-bossy’ approach to his students and his ability to listen and his
encouraging comments. It has always been interesting to listen to his comments either way. I feel both proud and privileged to be allowed to work in his lab and supervised by him.
Secondly, my heart felt thanks go to my co-supervisor Dr.Scient. Heidi Tveit. Her ability to stay optimist, her positive feedbacks, and her help and advice in designing experiments has been invaluable.
I would also like to thank all the members of the ‘PG’ group, former and current, for their help and for creating good scientific and social environment. The technical support of Truls Rasmussen is highly appreciable.
Special thanks to my late mother Tsehay Beka Gelalcha who has taught me to stand firm and strong in times of difficulties. I would like to thank Gifti, Sileshe, Salban, Deju, Torben, Duretti, Meseret, Serke, Betselot, Hailu Darge, Iskendir, Sosina, Rolf Henerik Sande, Olaf Øwre, Bergliot Øwre, Dejene Borana, Kumala Yadate, Moha, Tumsa, Barressa, Torben of Harvard, Abbaa Diimaa, Professor Girma of Upsala, Dr. Alemayehu for always believing in me and encouraging me throughout my study. Finally my sincere thanks to those not mentioned but by no means forgotten.
Last but not least, I would like to thank Marge and my children Hirre and Ayyantu for putting up with me. Hirre and Ayyantu, I was made for loving you!
5 Oslo, April 2011 Tilahun Tolesa Hafte
6
ABBREVIATIONS
AP Adaptor-protein complex COP Coat protein complex CS Chondroitin sulfate DS Dermatan sulfate ER Endoplasmatic reticulum
ERGIC ER-to-Golgi intermediate compartment GAG Glycosaminoglycan
Gal Galactose
GalNAc N-acetylgalactosamine
GalNAcT N-acetylgalactosaminyltransferase Gal T Galactosyltransferase
GFP Green fluorescent protein
GPI-Aps Glycosylphosphatydylinositol-anchored proteins GlcA Glucuronic acid
GlcNAc N-acetylglucosamine
GnT I/II N-acetylglucosaminyltransferase I/II
HS Heparan sulfate
IdoA Iduronic acid KS Keratan sulfate
7 LDLR Low density lipoprotein receptor ManI/II Mannosidase I/II
MDCK Madin-Darby canine kidney MS Mass spectrometry
NDST N-deacetylase-N-sulfotransferase NeuAc N-acetylneuraminic acid NeuGc N-glycolylneuraminic acid
PAPS 3’-phosphoadenosine-5’-phosphosulfate PAPST1 PAPS transporter 1
PC Procollagen
PG Proteoglycan
rGH Rat growth hormone
Ser Serine
SG Serglycin
TfR Transferrin receptor TGN Trans-Golgi network
Thr Threonine
UDP Uridine diphosphate
VIP36 Vesicular integral membrane protein of 36 kDa
Xyl Xylose
8
LIST OF PAPERS
Paper I
Protein core dependent glycosaminoglycan modification and glycosaminoglycan dependent polarized sorting in epithelial MDCK cells.
Hafte TT, Fagereng GL, Prydz K, Grøndahl F, and Tveit H.
Glycobiology vol. 21 no. 4 pp. 457–466, 2011
Paper II
N-glycan synthesis in the apical and basolateral pathway of epithelial MDCK cells on a model protein core and the influence of an additional glycosaminoglycan domain.
Moen A, Hafte TT, Tveit H, Egge-Jacobsen W, Prydz K.
Accepted manuscript (Glycobiology) with minor revisions Paper III
The protein conformation of the proteoglycan Serglycin influences the length and sulfation of the GAG chains.
Fagereng GL, Hafte TT, Tveit H, Prydz K Manuscript
9 Summary
Sorting of newly synthesized proteins in the secretory pathway of polarized cells is governed by signals of proteinacous nature or post-translationally added structures such as glycans.
Among all polarized cell types, epithelial cells are the most studied when it comes to polarized protein sorting, due to the differential access to the apical and basolateral
membrane domains obtained when cells are grown on permeable supports. The most studied epithelial cell line in culture is the Madin-Darby canine kidney (MDCK) cell line, isolated from a dog kidney cortex in 1958.
In the textbook view, proteins destined for the apical and basolateral membrane domains travel a common route through the secretory pathway to the trans-Golgi network, the exit site of the Golgi apparatus, from where further transport is mediated by specialized apical and basolateral transport containers. Some data obtained from studies of the soluble proteoglycan (PG) serglycin (Tveit et al. 2005; Vuong et al. 2006) and N-linked
glycoproteins (Alfalah et al. 2005), however, indicate that sorting may take place early in the secretory pathway. Serglycin was modified differently in the apical and basolateral secretory routes, where basolateral passage resulted in secreted serglycin which was much more intensely sulfated than the apically secreted variant, the latter accounting for 85 % of the secreted molecules (Tveit et al. 2005). Based on these data, we wanted to study further the apical sorting capability of the serglycin molecule and the differential post-translational modification in the two pathways.
Serglycin had been expressed with a green fluorescent protein (GFP) tag at the C-terminus, and for comparison, we chose to make all new variants and model proteins as GFP fusion proteins when expressed in MDCK cells. The glycosaminoglycan (GAG) attachment domain of serglycin was transferred to the junction between GFP and rat growth hormone (rGH), a non-glycosylated, non-sorted secretory protein. A variant of rGH carrying two N-
glycosylation sites in addition was also expressed, since these have been shown to mediate apical sorting in MDCK cells (Scheiffele et al. 1995). The serglycin GAG domain was mainly modified with chondroitin sulfate (CS) chains and mediated apical sorting in the rGH
10
context, but the higher sulfation intensity in the basolateral route was lost (Paper I). In fact, this capability seems to be coded by structures outside the GAG binding domain of serglycin, although not the internal disulfide bridge of the protein core (Paper III). The protein core also harbours a second type of apical sorting information of yet unidentified nature, since serglycin molecules from which all GAG attachment sites had been removed were also secreted mainly to the apical medium (paper III).
We also wanted to address whether N-glycans also could be differentially processed in the apical and basolateral secretory routes of MDCK cells. To this end, we expressed rGH with two sites for N-glycosylation and a C-terminal GFP tag, and a variant with the GAG binding domain of serglycin fused between rGH and GFP in addition. The N-glycan modifications were analysed by mass spectrometry. Only marginal differences in N-glycan processing and site occupancy were observed for the model proteins after their respective apical and basolateral secretion. Insertion of the GAG attachment domain, however, influenced the synthesis of the N-glycan structures in a more acidic direction. Thus in PGs that also carry N-glycans, the GAG modification may have an impact on N-glycan structure.
11
1. Introduction
1.1 Epithelial cells
The plasma membrane of epithelial cells is constituted by two domains, the apical and the basolateral, which are functionally distinct and exposed to different environments. Each domain has a characteristic lipid and protein composition. The apical surface is in contact with the external environment through invaginations of the body cavities, such as the lumen of the intestine, whereas the basolateral side faces the blood circulation system. These two domains are separated by “tight junctions”, a protein complex that seals adjacent epithelial cells and contributes to the maintenance of epithelial polarity by preventing lateral diffusion of molecules between the apical and basolateral membrane domains. Epithelial polarity is preserved by intracellular sorting mechanisms that maintain the different composition of lipids and proteins in the apical and basolateral surface domains (Matter 2000; Nelson and Yeaman 2001). Epithelial cells are easily grown on filters with differential access to the apical and basolateral domains, due to the fence function of the tight junctions. In this respect, cell polarity aspects have been much more extensively studied in epithelial cells, than in other polarized cells, like for instance neurons.
12
Fig. 1 A schematic representation of an epithelial cell with tight junctions.
1.1.1 MDCK cells
The Madin-Darby canine kidney (MDCK) cell line is an epithelial cell line isolated from dog kidney cortex and provides an excellent model system to study epithelial cell polarity. In fact, it is the most commonly studied epithelial cell line which, when grown on permeable supports, forms well polarized monolayers that can be accessed from both the apical and basolateral surfaces (Zegers and Hoekstra 1998). The MDCK cell line has been shown to retain many of the differentiated properties associated with kidney tubule epithelial cells (Herzlinger et al. 1982; Rindler et al. 1979; Valentich 1981). There are two main strains of MDCK cells termed MDCK I and MDCK II with the former resembling distal tubule epithelial cells and the latter sharing some characteristics with proximal tubule cells.
Another physiological difference between the two strains is the trans-epithelial electrical resistance. The MDCK I strain is characterized by having a higher electrical resistance
Tight junction Apical plasma membrane
ER
Nucleus ERGIC
Cis-Golgi Medial-Golgi Trans-Golgi Network (TGN)
Golgi Apparatus
Basolateral membrane
13
across filter-grown cell monolayers as compared to the MDCK II strain, and thus, MDCK I cells form tighter epithelia (Balcarova-Stander et al. 1984; Richardson et al. 1981).
1.2 The secretory pathway
1.2.1 Transport from the ER to the Golgi apparatus
Newly synthesized proteins destined for cellular compartments such as the plasma membrane, endosomes and lysosomes, utilize the secretory pathway to reach their final destinations. In this pathway, proteins travel from the endoplasmic reticulum (ER) through the Golgi apparatus to the plasma membrane, or to endosomes and lysosomes. The transport of proteins between membrane compartments of the secretory pathway takes place in a sequential manner involving both coated vesicular and tubular carriers. Secretory proteins must be sorted from residential ones and become enriched in transport carriers before undergoing the next step. All proteins in this pathway, whether resident or in transit, originate in the ER lumen. Secretory proteins are transported to their destination, either constitutively or in a regulated manner after translocation to the lumen of the ER. Such translocation is normally guided by a signal peptide at the N-terminus of the protein, but internal signals also exist, particularly for membrane proteins. The ER possesses quality control mechanisms that ensure the correct folding and assembly of both its own resident proteins, and proteins destined for other cellular locations. The protein folding process is aided by several chaperone proteins that assist malfolded proteins through several cycles of quality control. Finally, irreversibly misfolded proteins are targeted for degradation. Protein transport in the secretory pathway is a multistep process involving the generation of
transport carriers loaded with enrichment of defined sets of cargo molecules, the shipment of cargo-loaded transport carriers between compartments, and the specific fusion of these carriers with a target membrane (Derby and Gleeson 2007). The process of protein transport between the ER and the Golgi apparatus involves events like the collection of cargo molecules from the lumen of the ER followed by the formation of transport carriers that bud from the ER membrane and are transported in the direction of the Golgi apparatus. Protein trafficking in the secretory pathway can occur both via vesicle budding with subsequent
14
fusion with a target membrane and via passage along tubules that might connect to acceptor compartments (Vitale and Denecke 1999). ER exit sites (ERES), also known as transitional ER (tER), are specialized domains of ER membranes responsible for the directed export of secretory cargo (Palade 1975). The formation and budding of cargo vesicles at these ER sites involve the COPII coat protein complex that binds receptor protein tails at the cytoplasmic side of the membrane. The COPII complex thus mediates indirectly the selection of soluble cargo within the ER lumen; and transport from the ER towards the Golgi apparatus is initially mediated by vesicles coated with COPII coat proteins (Blazquez and Shennan 2000;
Watson and Stephens 2005). Proteins moving from the ER to the Golgi apparatus face another sorting station called the ER-Golgi intermediate compartment (ERGIC). ERGIC was originally defined by the mannose-binding transmembrane protein lectin ERGIC-53
(Appenzeller-Herzog and Hauri 2006) that cycles between the ER and the ERGIC, serving as a cargo receptor (Fiedler and Simons 1994; Itin et al. 1996). Glycoproteins, including cathepsin Z, cathepsin C, and the blood coagulation factors V and VIII were inefficiently secreted when ERGIC-53 was either mislocated to the ER or non-functional (Nichols et al.
1998; Vollenweider et al. 1998). In vitro analysis has demonstrated that the short cytoplasmic domain of ERGIC-53 contains binding sites for both COPI and COPII coat proteins, whereas the lumenal domain binds to immobilized mannose in a Ca2+ dependent manner (Itin et al. 1996). ERGIC is composed of several tubulovesicular membrane clusters proposed to constitute an obligatory membrane entity (Hauri et al. 2000) which is involved in both anterograde and retrograde transport between the Golgi apparatus and the ER.
ERGIC employs the COPI protein complex for its retrograde transport (Letourneur et al.
1994) activity and vesicles derived by COPII protein complex mediated budding
(Appenzeller et al. 1999) to receive cargo vesicles from the ER, thereby making the ERGIC the first post- ER sorting station for both anterograde transport and retrograde return of proteins. Further anterograde transport from the ERGIC to the cis-Golgi and through the Golgi apparatus does not seem to directly require COP proteins, but little is known about anterograde sorting from the ERGIC in the direction of the Golgi apparatus (Appenzeller- Herzog and Hauri 2006). Proper tethering of COPI vesicles to Golgi cisternae is, however, required for normal Golgi glycosylation (Smith and Lupashin 2008).
15
1.2.2 The Golgi apparatus
The Golgi apparatus or Golgi complex was discovered in 1898 by the Italian scientist Camillo Golgi, but was not until the 1970-ies shown to play a central role in the secretory pathway. It is primarily involved in the concentration, modification and sorting of newly synthesized protein and lipid molecules underway from the ER to the cell surface and endomembrane organelles. The Golgi complex consists of a series of membrane limited compartments through which proteins destined for the plasma membrane, secretory vesicles, endosomes and lysosomes move sequentially (Griffiths and Simons 1986). The central part of the Golgi apparatus may be observed as stacks of flattened cisternae which are classified as cis, medial, and trans, based on their functional proximity to the ER and plasma
membranes. The cis cisternae are enriched in the Golgi enzymes first encountered by cargo molecules (Bejarano et al. 2006; Prydz et al. 2008). The appearance of the Golgi apparatus differs significantly among cell type. In plants and lower animals, the Golgi apparatus exists as several copies of discrete stacks dispersed throughout the cytoplasm, while in vertebrate cells the Golgi stacks are normally localized in proximity to the nucleus (Shorter and Warren 2002). During protein trafficking, the Golgi apparatus, which is both dynamic and polar, receives cargo molecules from the ER via ERGIC at its cis-end and transports these from the cis end of the Golgi, through the medial-Golgi to the trans-Golgi region. The cis-Golgi functions to receive most of the biosynthetic output from the ERGIC, whereas the trans- Golgi network (TGN) sorts completed post-translationally modified products to their final destination (Shorter and Warren 2002).
1.2.3 Protein modifications in the Golgi apparatus
The Golgi apparatus is involved in post-translational protein modification mechanisms like glycosylation and proteolytic cleavage. Protein modification by glycans is initiated in the ER and proteins that arrive at the Golgi complex may be subjected to further post-translational modifications, such as removal of glycan units by glycosidases, additional glycosylation mechanisms, sulfation and proteolysis by the action of various Golgi resident enzymes in different cisternae. Golgi enzymes and associated substrate transporters are non-uniformly distributed within a Golgi stack, allowing sequential modification of cargo molecules in
16
transit (Smith and Lupashin 2008). Those Golgi processing enzymes which act early in glycan processing are usually concentrated in cis-Golgi cisternae, while those which act at later stages are confined to the trans-Golgi network (Farquhar and Palade 1981; Kweon et al. 2004; Martinez-Menarguez et al. 2001). N-acetylglucosaminyltransferase I (GlcNAcTI) and mannosidase II (ManII) are preferentially localized in cis/medial cisternae, while galactosyltransferases (GalT) and sialyltransferases (SiaT) are mostly found in the trans- Golgi region (Nilsson et al. 1993; Rabouille et al. 1995).
1.2.4 Protein journey through the Golgi apparatus
Most proteins arriving at the Golgi apparatus from the ER are glycosylated (N-
glycosylation) and are subjected to sequential modifications as they pass through the cis-, medial, and trans-Golgi cisternae. However, the mechanisms by which proteins travel through the Golgi stacks remain controversial among scientists in the field. Because of the structural complexity and highly dynamic nature of the Golgi apparatus, understanding the mechanisms that regulate cargo modification and trafficking through the cisternae has been elusive (Jackson 2009). The anterograde vesicular transport model and the cisternal maturation model are two alternative descriptions of intra-Golgi transport (Pelham and Rothman 2000). Previously, the Golgi apparatus was assumed to be a rather static membrane system, and based on this view, a vesicular transport model was proposed. According to this model, intra-Golgi trafficking is mediated through vesicles that bud from one cisternal compartment, which has its own defined environment and enzymatic content, followed by targeting to and fusion with the next acceptor compartment with cargo being transported from one face of the Golgi apparatus to the other in a series of vesicular transport steps (Jackson 2009; Mironov et al. 2005; Palade 1975; Simon 2008) in a COPI coat mediated manner (Lee et al. 2004). However, the vesicular transport model was challenged when it failed to explain the intra-Golgi transport of large proteins like procollagen-I (PC) which can move through the Golgi apparatus without entering COPI vesicles (Bonfanti et al. 1998;
Mironov et al. 2001). PC folds in the ER into rod-like trimers that further assemble in the Golgi into150 nm stable, cylindrical aggregates (Beck et al. 1996; Leblond 1989). PC- I aggregates, which are larger than COPI vesicles (50-60 nm diameter) traverse the Golgi stacks without ever leaving the lumen of Golgi cisternae (Bonfanti et al. 1998), supporting a
17
cisternal maturation model, which views the Golgi apparatus as a dynamic structure where the cisternae progress through the stack in the cis to trans direction with Golgi enzymes being recycled in a COPI vesicle dependent manner (Glick et al. 1997; Glick and Malhotra 1998). Thus, in both models, COPI vesicles are involved in intra-Golgi trafficking and maintenance of the normal structure of Golgi complex (Duden 2003). In the cisternal maturation model, cis-Golgi cisternal membranes are formed continuously by the fusion of pre-Golgi intermediates, an existing cis-cisterna becomes a medial one, and an existing medial cisterna becomes a trans cisterna (Glick 2000; Warren and Malhotra 1998). A basic difference between these two models lies in the differential movement of Golgi resident and cargo proteins, as indicated above. The vesicular transport model implies that the Golgi apparatus is a stable entity, where resident proteins are retained in the compartment while small vesicles mediate transport of cargo between subsequent cisternae along the Golgi stacks. Whereas, the cisternal maturation model predicts that cargo molecules remain within the lumen of the cisternae and move passively as cisternae mature and progress through the stack, while resident proteins are recycled by retrograde transport to a less mature cisterna, to maintain differential concentration across the stack in a COP I coated vesicle manner (Mironov et al. 2005; Smith and Lupashin 2008). Since the cisternal maturation model implies that Golgi enzymes are concentrated in COPI coated vesicles, has this possibility been investigated (Lanoix et al. 1999; Lanoix et al. 2001; Martinez-Menarguez et al. 2001).
Although data supporting the view that Golgi enzymes are enriched in COPI vesicles was provided, this model was challenged by the finding of mannnosidase II deficient Golgi associated COPI vesicles (Cosson et al. 2002; Orci et al. 2000). The mannosidase II depleted COPI vesicles were enriched in KDEL receptors, which bind the ER retention sequence KDEL, suggesting the function of COPI coat vesicles to be restricted to transport from Golgi cisternae and ERGIC back to the ER. The role of COPI in retrograde transport in the maturation model was also questioned when COPI vesicles containing anterograde cargo proteins like pro-insulin (Orci et al. 1997) were identified, opposed to the assumption of the maturation model that cargo molecules do not leave the lumen of the cisternae. The contents and directionality of COPI vesicles are thus not entirely clear, but at present the maturation model gains most support from authorities in the field (Emr et al. 2009). The observation of tubular connections between cisternae of the Golgi complex (Marsh et al. 2004; Trucco et al.
2004) led to the suggestion of a third model called the continuity- based model. This model
18
predicts that different Golgi cisternae are connected with each other through tubular structures which could facilitate the flow of large cargo like PC in a cis-to the-trans direction, as well as the retrograde transport of Golgi enzymes in a COPI vesicles
independent manner (Mironov et al. 2005). The fourth alternative model that has emerged, the rapid-partitioning model, postulates the movement of both Golgi resident enzymes and cargo molecules in both the anterograde and retrograde directions through the Golgi apparatus (Jackson 2009; Patterson et al. 2008) indicating a clear departure from the cisternal maturation model, which asserts actively sorted trafficking of Golgi enzymes only in the trans-to-cis direction. However, a unification of the anterograde vesicle transport model and the cisternal maturation model, the bidirectional trafficking of cargo molecules within the Golgi apparatus has been suggested previously (Pelham and Rothman 2000).
Another interesting feature of the rapid partitioning model is the inclusion of a lipid sorting concept in the trafficking pathways, where cargo and glycosylation enzymes have their own lipids of preference to associate with in different Golgi domains. Golgi domains enriched in glycerophospholipids (GPLs) usually accommodate Golgi processing enzymes while Golgi domains with high content of sphingolipids (SLs) harbour cargo molecules (Jackson 2009).
19
Fig.2. Two main models for intra-Golgi trafficking. Cargo synthesized in the ER and transported through the secretory pathway is shown in yellow; Golgi processing enzymes are shown in blue. Golgi element 1 is at the cisside and receives material from the ER, and Golgi element 4 is at the transside and is involved in packaging of cargo for delivery to the plasma membrane (PM). Arrows indicate the direction of trafficking. (A) Forward vesicular- trafficking model. Vesicles carrying cargo bud from a donor compartment, and are then targeted to and fuse with the following compartment in the secretory pathway (acceptor compartment). (B) Cisternal-maturation model. Vesicles carrying Golgi processing enzymes bud from a later compartment in the secretory pathway (donor compartment), then fuse with an earlier compartment (acceptor compartment). The cargo progresses as a result of maturation of an earlier compartment into a later one. Adapted from Jackson, C.L.(2009).
1.2.5 Sorting of proteins in the secretory pathway of epithelial cells
In the classical view of the Golgi apparatus, the TGN plays an important role in directing secretory proteins to their appropriate destinations, by serving as a site for the sorting of proteins to various cellular components such as endosomes, lysosomes, and different domains of the plasma membrane. The TGN is not only involved in the sorting of secretory proteins to different destinations, but also in receiving extracellular material and recycled molecules from endocytic compartments (Gu et al. 2001). In addition, proteins can be routed from the TGN to their final target site through different pathways. Direct pathways deliver proteins to the cell surface, while indirect pathways sort proteins to endosomes for later transport to the plasma membrane (Ang et al. 2004). In epithelial MDCK cells, pathways for direct delivery of newly synthesized proteins exist to both the apical and basolateral surfaces (Jacob and Naim 2001; Kreitzer et al. 2003), ensuring the asymmetric distribution of proteins that require sorting (Traub and Kornfeld 1997). In other epithelia, indirect pathways are common, where proteins are first transported to the basolateral surface domain from where they are endocytosed and delivered to the opposite surface via early and recycling endosomes (Huet et al. 2003; Mostov et al. 2000). However, the mechanisms behind the sorting of proteins directly from the TGN to the PM or into the indirect transendosomal route to the PM are not fully elucidated (Gravotta et al. 2007).
20
1.3 Sorting signals for the journey beyond the Golgi apparatus
The sorting machinery in the TGN controls multiple divergent pathways directed to spatially segregated acceptor compartments, such as the apical and basolateral plasma membranes of epithelial cells, early/sorting endosomes or late endosomes, recycling endosomes, secretory granules or Golgi stacks accepting retrogradely transported molecules (De Matteis and Luini 2008). A wide variety of signals located within the cytoplasmic domains of transmembrane proteins can mediate exit from the Golgi complex, including signals (DXXLL motif based) for sorting of mannose-6-phosphate receptors (M6PRs) which target newly synthesized acidic hydrolases modified with mannose-6-phosphate to the lysosomal compartment (Bonifacino and Traub 2003), and signals targeting proteins to the apical and basolateral domains of the plasma membrane in epithelial cells (Rodriguez-Boulan et al. 2005).
Knowledge concerning the sorting signals directing secretory proteins in the apical and basolateral biosynthetic pathways is limited. However, N- and O- glycans, GPI-anchors, and glycosaminoglycan (GAG) chains, have been suggested to contain sorting information that drives secretory proteins in the apical direction (Alfalah et al. 1999; Benting et al. 1999;
Kolset et al. 1999b; Scheiffele et al. 1995).
1.3.1 Basolateral sorting signals
Polarized epithelial cells maintain their polarity by ensuring asymmetric distribution of newly synthesized proteins, transporting distinct populations of proteins to their apical and basolateral membrane domains in a signal- mediated manner. Much work has been carried out to identify the sorting signals that drive proteins along the secretory pathway. The first basolateral sorting signal to be proposed was a signal in the cytoplasmic tail of the polymeric immunoglobulin A receptor (pIgR) (Mostov et al. 1986). Basolateral sorting of
transmembrane protein is now generally known to depend on cytoplasmic peptide sequences, some of which conform to tyrosine ( NPXY or Yxx) consensus motifs and others to dileucine (DXXLL or [DE]XXXL[LI]) based motifs (Nelson and Yeaman 2001).
Additional discoveries, like a basolateral sorting signal for Stem Cell Factor (SCF) comprising a single leucine motif with a cluster of acidic amino acids at its N-terminal
21
(Wehrle-Haller and Imhof 2001) further substantiated the evidence that basolateral sorting of transmembrane proteins can be mediated by a variety of signals, and that the signal is located at the cytoplasmic domain. A glycosylphosphatidyl-inositol (GPI) linked heparan sulfate proteoglycan (HSPG), glypican, which when expressed in MDCK cells, appeared at the basolateral membrane was detected mostly at the apical membrane after removal of the site for HS GAG chain attachment (Mertens et al. 1996), suggesting a role of HS GAG chains in basolateral sorting of glypican. Basolateral signals with tyrosine and dileucine based motifs interact with adaptor protein (AP) complexes and clathrin to direct proteins to the basolateral cell surface (Bonifacino and Traub 2003; Potter et al. 2006b; Rodriguez- Boulan and Musch 2005). Several adaptor protein family members like AP1B, AP3 and AP4 seem to be involved in sorting of basolateral membrane proteins (Folsch et al. 1999; Simmen et al. 2002). AP1B and AP1A with common , , and subunits are involved in endosomal targeting from the TGN in a clathrin dependent manner (Ohno et al. 1999). However, AP1B, having an epithelial-specific μ1B subunit (Ohno et al. 1999) is also involved in promoting the sorting of LDL and transferrin receptors (LDLR and TfR) at recycling endosomes (Gan et al. 2002). The sorting of the Mannose-6-phosphate receptor (M6PR) at the TGN to endosomes, is facilitated by the cooperative activity of AP1 and GGAs (Golgi localized, - ear-containing, Arf binding proteins) which require the involvement of clathrin (Bonifacino and Lippincott-Schwartz 2003; Ghosh et al. 2003; Puertollano et al. 2001). AP3 has been shown to regulate the exit of vesicular stomatitis virus glycoprotein (VSVG) from the Golgi complex in non-polarized cells (Nishimura et al. 2002), while AP4 has been proposed to participate in the basolateral sorting of LDLR, TfR and M6PR (Simmen et al. 2002). In contrast to other adaptor proteins like AP1, AP2 and AP3 which have a predicted clathrin binding site, AP4 is devoid of such a clathrin binding site and might therefore participate in the formation of tubules directed to the basolateral membrane (Rodriguez-Boulan and Musch 2005). Clathrin has been implicated to have a role in the transport of PM protein to the basolateral membrane of epithelial MDCK cells. The depletion of clathrin in MDCK cells by siRNA led to loss of polarized transport of basolateral PM proteins including TfR, VSVG, E-cadherin, and neural cell adhesion molecule (NCAM) (Deborde et al. 2008). In contrast to the sorting signals of TM proteins in the basolateral pathway, sorting signals directing secretory proteins in the basolateral direction have not been determined to date.
However, for secretory proteins such as apoA-I and apoA-II, it has been suggested that
22
preferential basolateral transport is facilitated by cell-dependent default pathways and not by sorting signals in both Caco-2 and MDCK cells (Remaley and Hoeg 1995; Rindler and Traber 1988).
1.3.2 Apical sorting signals
In contrast to sorting signals discovered for basolateral transmembrane proteins, which are mainly confined to their cytoplasmic domain, apical sorting signals known to date are more diverse. Different types of apical sorting signals, including those located in extracellular, transmembrane, and cytoplasmic protein domains (Lin et al. 1998; Mostov et al. 2000) drive apically destined proteins along the biosynthetic route. In addition, apical sorting signals have also been postulated to reside in the glycosylphosphatidylinositol (GPI) lipid modification and N- and O-glycan moities of proteins (Potter et al. 2006). Apical sorting information might also be localized to the glycan moieties of chondroitin sulfate (CS) proteoglycans (Kolset et al. 1999). The utilization of diverse sorting signals for apical proteins makes the elucidation of the mechanisms behind the delivery in this direction difficult. The observation of the presence of glycosylphosphatidylinositol-anchored proteins (GPI-Aps) at the apical surface of MDCK cells (Lisanti et al. 1988) led to the postulation of the GPI-anchor as an apical sorting signal, because recombinant addition of a GPI-anchor to a secretory protein (Lisanti et al. 1989) resulted in apical surface localization of the chimeric plasma membrane protein (Rodriguez-Boulan et al. 2005). However, a selective point mutation resulting in removal of the GPI-anchor from naturally N-glycosylated and GPI- anchored membrane dipeptidase (MDP) did not affect apical sorting (Pang et al. 2004) and the coupling of the secretory protein rGH with a GPI- anchor did not increase apical transport of the protein (Benting et al. 1999). These data indicated that a GPI-anchor is not always sufficient to route a protein to the apical surface. The apical sorting of GPI-anchored proteins was suggested to be the result of the ability of such proteins to segregate into glycolipid- and cholesterol-enriched microdomains or lipid “rafts” (Brown and London 1998). The introduction of N-glycans in the secretory protein rat growth hormone (rGH) (Scheiffele et al. 1995) and the GPI-anchored version of rGH (Benting et al. 1999) in MDCK cells changed its surface transport pattern, triggering more apical secretion as compared to secretory and GPI-linked unglycosylated rGH. N-glycans on several secretory
23
proteins like erythropoetin (Epo) (Kitagawa et al. 1994) and endolyn (Potter et al. 2006), and on naturally N-glycosylated and GPI anchored membrane dipeptidase (MDP) (Pang et al. 2004) confer apical sorting in MDCK cells, N-glycans are, however, not universal signals for apical sorting. The complete removal of N-glycans from the soluble secretory protein hepatitis B surface antigen (HBsAg) (Marzolo et al. 1997), and from the voltage- and Ca2+- activated K+ channel alpha -subunit endogenous plasma membrane protein in MDCK cells (Bravo-Zehnder et al. 2000) did not affect their apical transport, suggesting that
oligosaccharide targeting is one of multiple mechanisms that can promote apical sorting (Mostov et al. 2000). Apical sorting can also be mediated by O-glycosylation (Weisz and Rodriguez-Boulan 2009). The addition of an O-glycosylated stalk region, which is important for apical sorting of the small intestinal membrane glycoprotein sucrase-isomaltase (SI), to the secretory protein rGH by converting it to a membrane protein, resulted in strong sorting preference to the apical surface (Alfalah et al. 1999; Spodsberg et al. 2001). O-glycosylation mediated the association of rGH with lipid rafts (Weisz and Rodriguez-Boulan 2009). Apical targeting information may also be constituted by proteinaceous motifs (Rodriguez-Boulan and Musch 2005). Sorting signals in cytoplasmic tail sequences of apical proteins, like the multiligand receptor megalin (Takeda et al. 2003), the light sensitive protein rhodopsin (Chuang and Sung 1998), and receptor guanylyl cyclase (GC) (Hodson et al. 2006) target these proteins to the apical membrane. Apical sorting signals can also reside within transmembrane domains of apically sorted proteins (Folsch et al. 2009) as shown for influenza haemagglutnin (HA) and neuraminidase (NA) (Kundu et al. 1996; Lin et al. 1998).
The identity of apical sorting signals is, however, a matter of continued debate (Ihrke et al.
2001), and the sorting mechanisms utilized by apical signals are still poorly understood (Rodriguez-Boulan and Musch 2005). Inhibition of glycosylation of the tight junction protein occludin resulted in accumulation of the protein in the Golgi complex of MDCK cells (Gut et al. 1998), implying the involvement of oligosaccharides in protein sorting. A suggested sorting lectin for apical selection of cargo was the vesicular integral protein (VIP) 36 (Fiedler and Simons 1996), proposed to recognize carbohydrates, such as N-glycans in the TGN. However, the discovery that VIP 36 is predominantly localized to the early secretory pathway (Fullekrug et al. 1999) and recognizes high-mannose glycans (Yamaguchi et al. 2007) , put the candidacy of this vesicular integral membrane protein as an apical sorting receptor to rest. Interestingly, the proposal that sorting of cargo in the biosynthetic
24
pathway of polarized cells could occur earlier than previously thought (Alfalah et al. 2005;
Tveit et al. 2005) might reinstate VIP 36 as a potential sorting receptor for apical secretion, but VIP36 might actually be involved in a quality control cycle between the ER and the Golgi apparatus (Reiterer et al. 2010). The importance of oligomerization for proper targeting and stabilization of GPI-Aps into rafts was also proposed from the observation that GPI-anchored green fluorescent protein (GPI-GFP) was misrouted to the basolateral surface when oligomerization/cluster formation was impaired by the introduction of a mutation in the GFP gene (Paladino et al. 2004).The lipid raft hypothesis which postulates that certain proteins, among these GPI-linked proteins, are sorted apically, because of their affinity for glycosphingolipid and cholesterol rich domains was challenged, however, when it failed to account for the presence of some GPI-Aps on the basolateral surface of MDCK cells (Paladino et al. 2004). However, further investigation revealed that the level of cholesterol was a driving factor for the proper oligomerization, which in turn was necessary for the proper apical sorting of GPI-APs. This was verified when addition of cholesterol resulted in oligomerization and more apical transport of GFP-PrP which is sorted basolaterally as monomers (Paladino et al. 2008), indicating the importance of the lipid environment in apical sorting.
25
2. Glycoproteins and proteoglycans
Glycoproteins are proteins to which carbohydrates are covalently attached and modified in a stepwise manner by a process called glycosylation. The attachment of oligosaccharides to a protein can occur either by co-translational or posttranslational mechanisms and is catalyzed by the action of a variety of glycosyltransferases. Glycans play important functional and structural roles in both membrane-associated and secreted proteins and can be classified according to their attachment site in a protein. Most glycoproteins carry N-glycans, where the carbohydrate is attached to a polypeptide through the amide group of an asparagine (Asn) residue forming N-glycosidic bonds. O-glycans are linked to a protein through the hydroxyl group of serine, threonine or hydroxylysine residues of polypeptides which result in the formation of O-glycosidic bonds. Proteoglycans (PGs) are glycoproteins consisting of a protein core to which chains of repeating disaccharide units called glycosaminoglycans (GAGs) are attached via a linker tetrasaccharide to the hydroxyl group of a serine residue.
GAG disaccharides are most often composed of either N-acetylglucosamine (GlcNAc) or N- acetylgalactosamine (GalNAc), and a uronic acid such as glucuronate or iduronate. GAGs are unbranched polysaccharides with high negative charge, due to modification of the disaccharide units by sulfate and the charge of the uronic acid groups. PGs can be classified as chondroitin sulfate (CS)/ dermatan sulfate (DS), heparin/heparan sulfate (HS), and keratan sulfate (KS) based on the type of GAGs they contain. The only GAG with no bound core protein is hyaluronic acid, which contains GlcNAc and GlcA sugars.
2.1 Synthesis of N- and O -linked Glycans
The synthesis of N-linked glycoproteins is initiated co-translationally in the endoplasmic reticulum (ER) and is completed in the Golgi apparatus by post-translational modifications involving a number of glycosidases and glycosyltransferases. An N-linked oligosaccharide arises when a core oligosaccharide (Glc3Man9GlcNAc2) assembled on the lipid carrier dolicholpyrophosphate in the ER membrane, is transferred cotranslationally to nascent polypeptide chains with the acceptor sequon Asn-X-Ser/Thr, where X can be any amino acid except proline (Freeze and Aebi 2005; Gavel and Vonheijne 1990; Helenius and Aebi 2001),
26
by the action of oligosaccharyltransferase (OST), localized to the ER membrane (Silberstein and Gilmore 1996). In the ER, protein-bound N-linked oligosaccharides have specific functions in protein folding and quality control, and in forward transport.
Fig. 3. Biosynthesis of N-linked glycans. Adapted from (Helenius and Aebi 2001)
The overall structure of N-glycans can be classified as complex, hybrid, or high mannoses (oligomannose), which share a common core sugar structure, Man -(1–6) - Man -(1–3)- Man -(1–4)-GlcNAc -(1–4)-GlcNAc 1-Asn-X-Ser/Thr. The three major stages in the pathway for synthesis of N-glycans include the assembly of the core oligosaccharide on the lipid carrier dolichol through a pyrophosphate linkage by the action of enzymes on both the cytoplasmic and the lumenal side of the ER (Gahmberg and Tolvanen 1996; Helenius and Aebi 2001; Kornfeld and Kornfeld 1985). The synthesis of the lipid-linked precursor starts at the cytoplasmic face of the ER membrane by the addition of two GlcNAc and five mannose residues provided by nucleotide sugar donors, UDP-GlcNAc and GDP-Man, respectively.
When this is completed, the lipid carrier with bound sugars is flipped to the ER lumen, where four more mannose and three glucose residues are provided as dolichol linked sugars which are added one sugar at the time to form a complete core oligosaccharide. The transfer of the completed lipid-linked core oligosaccharide to a growing nascent polypeptide
27
acceptor also takes place in the lumen of the ER and the core oligosaccharide is coupled to the asparagin residue in an Asn-X-Ser/Thr acceptor sequence catalyzed by
oligosaccharyltransferase, a complex enzyme with its active site in the lumen of the ER (Silberstein and Gilmore 1996), marking the second stage in the synthesis of N-glycans. The second phase of the glycosylation process involves both the removal of selected
monosaccharides in the ER and the rebuilding by addition of other monosaccharides in the Golgi apparatus. This introduces structural diversity to the newly synthesized N-glycans, due to non-uniform modifications, especially terminal glycosylation in the Golgi complex as a result of the variation in the expression and organization of modifying enzymes in the Golgi apparatus as well as the availability of substrates (Freeze and Aebi 2005). N-glycans have a common role in events like protein folding, quality control, and in sorting events early in the secretory pathway, whereas further modification in the Golgi gives rise to diverse mature glycoproteins on the cell surface (Helenius and Aebi 2001). N-glycans transferred to the nascent polypeptide are subjected to further trimming, where terminal glucose and mannose residues are removed sequentially by the action of glucosidases I and II, and mannosidase I in the ER (Kornfeld and Kornfeld 1985). Glucosidase I is responsible for the removal of the terminal glucose residue, whereas glucosidase II removes the remaining two inner glucose residues, signalling that the newly synthesized glycoprotein is ready to transit from the ER to the Golgi apparatus, provided that it has passed the quality control regime, which ensures the correct folding of newly synthesized glycoproteins, assisted by calnexin and calreticulin lectin chaperones in the ER. The removal of three terminal glucose residues is imperative in order that further processing to the mature carbohydrate unit can take place (Spiro 2000). ER and cis-Golgi resident mannosidases together remove four additional mannose residues from the core oligosaccharide, generating a series of oligomannose-type N-glycans in the ER and the Golgi apparatus (Tomiya et al.
2004) and these high mannose variants move through various compartments to the cell surface, escaping further modifications constituting the third stage in the synthesis of N- glycans. This stage primarily occurs in the Golgi apparatus, where synthesis of other complex glycan structures is initiated by the action of Golgi resident transferases. It starts when N-acetylglucosaminyltransferase I (GnT I ), which is located in the medial portion of the Golgi, transfers one GlcNAc residue donated by UDP-GlcNAc to the Man -(1-3) arm of the core oligosaccharide, followed by the subsequent removal of two mannose residues by
28
medial-Golgi mannosidase II from the (1-6) arm of the core oligosaccharide. N-
acetylglucosaminyltransferase II (GnT II) then adds the second GlcNAc residue to the (1-6) arm of the core, resulting in the synthesis of a bi-antennary complex structure (Schachter 2000). Further in the secretory pathway, the complex structures are extended by the addition of a single galactose unit donated by UDP-Gal and a sialic residue from cytidine-
monophosphate-NeuAc (CMP-NeuAc) to each terminal GlcNAc residue in the core oligosaccharide, by the action of galactosyltransferases (GalTs) and sialyltransferases (SiaTs), respectively in the trans-Golgi compartment. The activated donor nucleotide sugars are made in the cytoplasm and must be transported into the lumenal compartment of the ER, facilitated by nucleotide sugar transporter molecules (Freeze and Aebi 2005). N-glycans with hybrid structure are formed when -mannosidase no longer acts after the introduction of the first GlcNAc residue to the core oligosaccharide by GnT I, due to the competition by other glycosyltransferases for the common substrate (Schachter 1991). Different cell types can generate a variety of biosynthetic enzymes which can transfer different sugar residues to a growing oligosaccharide group, thereby synthesizing glycans of variable structure on a given protein in a cell type and tissue specific manner. These structures can undergo time- dependent changes during development and metastatic progression (Schachter 1991).
Altered expression of GnT-V was suggested to affect the glycan structure and function of epidermal growth factor receptor (EGFR) in hepatocarcinoma cells (Guo et al. 2004). Mouse embryos lacking GnT I, producing only the high-mannose type, was shown to have failure in neural tube formation, vascular system malformation, and lack of bronchial epithelium (Fukuda and Akama 2003). Altered glycosylation in terms of alteration in sialylation and fucosylation of the cell surface exposed glycans have been observed in MDCK cells after transformation by murine sarcoma virus (Bruyneel et al. 1990). Altered glycosylation might also be caused by a change in the pH of the Golgi apparatus lumen which impairs
glycosylation by interfering in processes including nucleotide sugar transport, activity of glycosyltransferases in either reducing their activity or mislocalizing of Golgi
glycosyltransferases into post-Golgi organelles (Axelsson et al. 2001; Gawlitzek et al. 2000;
Waldman and Rudnick 1990). Elevated Golgi pH affected the status of terminal N- glycosylation in African green monkey kidney cells (COS-7) (Rivinoja et al. 2009).
However, the mechanism by which neutralization of the slightly acidic pH in the Golgi affects glycosylation is not known. Most of Aberrant glycosylations are results of altered
29
SiT expression in human cancers (Sotiropoulou et al. 2002; Wang et al. 2005). Increased fucosylation and /or decreased sialylation has been observed in cystic fibrosis airway epithelial cells as compared with the non-cystic fibrosis, due to a change in the activity of fucosyltransferases (Glick et al. 2001). Increased expressions of -(2-6)-linked sialic acids on N-glycans have been usually associated with human cancer progression, metastatic spread (Hedlund et al. 2008). Mass spectrometric analysis of glycan structures in Swine Respiratory Epithelial cells (SRECs), a primary target cell type of influenza virus, revealed the presence of a variety of glycans terminating in sialic acid, where -(2-6)-linked sialylation was more abundant than -(2-3)-linked sialylation, and N-acetylneuraminic acid (NeuAc) was more abundant than N-glycolylneuraminic acid (NeuGc) (Bateman et al.
2010). Both human respiratory tract and chicken intestine express NeuAc, whereas both NeuAc and NeuGc are present in swine trachea (Guo et al. 2007; Suzuki et al. 1997). Neural Cell Adhesion Molecules (NCAMs), which play a role in neural development, are highly polysialylated in embryonic tissues, as compared with adult tissues (Angata et al. 2000; Kiss and Rougon 1997; Rutishauser and Landmesser 1996) indicating the variation of the N- glycan structure with time, tissue and developmental stage. The abundance of polysialylated NCAM molecules is dependent on the expression level of different polysialtransferases during different developmental stages (Angata and Fukuda 2003; Muhlenhoff et al. 1998).
The structure of N-glycans, both secretory and transmembrane, derived from MDCK cell is not well studied. The structural analysis of N-glycans, both secretory and transmembrane variants, derived from the apical and basolateral membrane domains of polarized MDCK cells has revealed no significant difference in the overall structure of N-glycans between the two domains. However, the secretory proteins from both the apical and basolateral domains were predominantly composed of N-glycans with Sia -(2-6)-GalNAc -(1- 4)-GlcNAc sequence (Ohkura et al. 2002). N-glycans with high-mannose structure have also been observed to appear at the cell surface, possibly utilizing a secretory transport route bypassing the Golgi apparatus. For instance, protein-tyrosine phosphatase CD45 in T lymphoma cells (Baldwin and Ostergaard 2002), the AE1 anion exchanger (Ghosh et al. 1999), and the subunits of the sodium channel (ENaC) at the apical surface of epithelial cells (Weisz and Johnson 2003), predominantly contain high-mannose sugars which are endo-H sensitive.
Defects in the N-glycan synthesis pathway have been implicated in different types of congenital disorders of glycosylation (CDG) which result from deficiencies in either the
30
biosynthesis of oligosaccharide precursors or steps in N-glycan assembly, affecting especially normal brain development and functions of the nervous, hepatic, gastrointestinal and immune systems (Freeze and Aebi 2005; Jaeken 2003; Marquardt and Denecke 2003).
Fig. 4 Main types of N-glycans. Modified from “Essentials of Glycobiology” second edition (2009).
O-linked glycans are usually added to a protein core through the hydroxyl group of Ser or Thr residues on the polypeptide in the Golgi apparatus. In contrast to the synthesis of N- glycans, addition of O-glycans to a polypeptide does not require a lipid carrier in eukaryotes O-linked glycoproteins are synthesized by the stepwise addition of one sugar at the time, donated from nucleotide sugars imported into the secretory pathway, directly onto the glycoprotein by a glycosyltransferase. The first step in the O-glycosylation pathway is transfer of GalNAc to either a serine or threonine residue of a protein (Potapenko et al.
2010), by the action of a N-acetylgalactosaminyltransferase (Ten Hagen et al. 2003). There is no known consensus amino acid sequence or core structure identified for O-glycosylation, it can occur on several adjacent Ser or Thr residues, especially in mucins (Gill et al. 2010).
O-linked glycans have been implicated in a variety of biological activities, including as
31
recognition markers for different blood antigens, as adhesive ligands, and as modulators of cell-cell signalling (Aoki et al. 2008).
2.2 Synthesis of proteoglycans (PGs) and glycosaminoglycans (GAGs)
PGs are found at the cell surface, in the extracellular matrix (ECM) or intracellularly in secretory granules (Kjellen and Lindahl 1991; Kolset et al. 2004).The biological functions of GAGs range from functions such as modulating signal transduction pathways, cell proliferation and angiogenesis, adhesion, cell migration (Souza-Fernandes et al. 2006), wound healing and blood clotting (Kuberan et al. 2008). PGs can also be used by many pathogens for entry into cells (Handel et al. 2005). GAGs interact with specific proteins depending on their sulfation pattern. Particular sulfation patterns in heparan sulfate GAG chains allow interactions of ionic nature with growth factors (Prydz and Dalen 2000;
Schaefer and Schaefer 2010). The secretory pathway is the site of initiation, formation of the GAG linker region. Further elongation/polymerization and modification of the GAG chains involves various biosynthetic enzymes. Activated sugar and sulfate donors like UDP-sugars and 3’-phosphoadenosine 5’-phosphosulpate (PAPS) are required for the biosynthesis of GAGs. These activated sugars and sulfate donors are then translocated from the cytoplasm to the lumen of the ER and the Golgi apparatus by specific sugar and sulfate transporters (Hirschberg et al. 1998; Mandon et al. 1994). The formation of GAGs occurs through a sequence of events. The first step in GAG biosynthesis is chain initiation by the transfer of a xylose (xyl) residue to a serine amino acid adjacent to a glycine in the core protein, followed by the assembly of a tetrasaccharide linkage serving as acceptor for the chain elongation by the alternate addition of D-gluronic acid and aminosugars (Garud et al. 2008; Kuberan et al.
2008; Prydz and Dalen 2000; Victor et al. 2009). In the linker tetrasaccharide the xyl is followed by two galactoses (Gal) and finally glucuronic acid (GlcA) forming a common linkage [GlcA -(1–3)-Gal -(1–3) Gal -(1–4)Xyl -(1-O-Ser)] between the core protein and the GAG chain (Lindahl and Roden 1966). The synthesis of the linker tetrasaccharide is probably initiated in the ER or a subsequent compartment by the addition of xylose to a core protein, followed by addition of the two galactoses in the cis/medial Golgi region and the
32
linker is completed by the addition of the glucuronic acid unit in the medial/trans Golgi (Silbert and Sugumaran 2002). The linker region can be modified by phosphorylation of xylose at C-2 for both CS and HS GAGs (Fransson et al. 1985; Oegema et al. 1984).
Sulfation of one or both galactose residues in the linker region of CS/DS has been found in various tissues, but never in HS/heparin (Sugahara and Kitagawa 2000). It is the addition of the fifth sugar that decides whether a GAG chain becomes glucosaminoglycan (HS/heparin) or galactosaminoglycan (CS/DS), since the same enzymes are involved in the biosynthesis of the linker tetrasaccharide for both types of GAG chains (Kjellen and Lindahl 1991;
Prydz and Dalen 2000). The mechanism that commits the process towards the synthesis of CS/DS or HS/heparin PGs is not fully understood, and a single consensus sequence for chain initiation is not clearly defined yet. However, the GAG binding domain serving as an acceptor site for xyl transfer has a repetitive Ser-Gly consensus sequence flanked by a cluster of acidic amino acid residues for the synthesis of HS in betaglycan and the substitution of these acidic residues resulted in the synthesis of more CS (Zhang and Esko 1994). This might indicate the importance of the GAG binding domain and the acidic residues nearby in determining the outcome of the GAG synthesis process. The mutation of an adjacent tryptophan residue resulted in the synthesis of less HS and a shift towards the synthesis of more CS, while the introduction of a tryptophan closer to a CS site triggered the production of more HS (Zhang and Esko 1994), again implying the influence a globular residue might have on the type of GAG chains to be synthesized. Recent studies of
alignment of amino acid sequences of several CS attachment sites from various core proteins resulted in the generation of the consensus sequence a-a-a-a-Gly-Ser-Gly-a-b-a (a =Glu/Asp and b = Gly, Glu or Asp) (Silbert and Sugumaran 2002). Serglycin isolated from the rat yolk sac carcinoma cell line L2, on the other hand contains 24 consecutive Ser-Gly repeats where only one of these repeats is flanked by acidic amino acid residues (Avraham et al. 1989;
Bourdon et al. 1986; Kjellen et al. 1989; Stevens et al. 1988). All these findings, however, do not indicate the existence of a universal consensus sequence for GAG attachment onto the core protein. The type of GAG chain synthesized might also be affected by a domain at a distance from the GAG binding sites in the protein core. Insertion of an N-terminal globular domain of the HSPG glypican-1 at a significant distance from the GAG binding sites in another PG enhanced the assembly of HS chains, while removal of the same globular domain from glypican-1 converted this PG into a PG with predominantly CS chains (Chen
33
and Lander 2001). Cellular mechanisms underlying GAG biosynthesis have also been studied using different exogenous xylosides called click-xylosides, containing various hydrophobic aglycone moieties in Chinese hamster ovary (CHO) cells. The study revealed that the introduction of external click-xylosides generated free GAG chains of variable CS composition and quantity, as well as GAG chains with different sulfation intensity, sulfation pattern and chain length (Victor et al. 2009). The synthesis of GAG chains with different modifications after the admnistration of xylosides to CHO cells can be explained in light of the GAGosome model. In this model, each GAGosome is composed of different set of modifying enzymes which, when recruited, generate heterogeneous GAG chains with variable structure, sulfation pattern as well as length. In contrast to heparin that occurs exclusively in the connective tissue mast cells, HS is produced by most cells in the body (Lindahl and Li 2009). CS/DS PGs, on the other hand, are synthesized mostly in vertebrate cells as major components of connective tissue matrix (Kjellen and Lindahl 1991b; Lindahl and Li 2009).
Keratan sulfate (KS) is another sulfated GAG and is abundant in cornea, and is also present in cartilage and brain. KS does not contain the same linker tetrasaccharide structure as CS- and HSPGs. The core structure of KS consists of repeating -[3Gal-1-4GlcNAc-1]- disaccharide units (N-acetyllactosamine) attached to a complex type biantennary
oligosaccharide N-linked on Asn residue in KS type I in cornea, whereas the core structure of KS II in cartilage is attached to protein through GalNAc-O-Ser/Thr (Funderburgh 2000).
KS is synthesized via the action of glycosyltransferases that alternately add Gal and GlcNAc residues to the growing polymer. In humans, 4GalT-IV and 3Gn-T7 have been shown to activate the polymerization of oligosaccharides in corneal KS (Kitayama et al. 2007). The core oligosaccharide polymer in KS can be modified by sulfation occuring on C-6 of both Gal and GlcNAc residues (Lauder et al. 1997). Sulfate modification of KS GAGs varies between tissues and is carried out by a variety of sulfotransferases (STs). For instance, 6-O- sulfation of GlcNAc residues is carried out by GlcNAc-6ST-5 in cornea (Akama et al. 2000;
Hayashida et al. 2006), whereas GlcNAc-6ST-1 catalyzes the sulfation of KS GAGs in brain (Zhang et al. 2006).
34
Fig. 5. Schematic structure of glycosaminoglycan chains attached to a proteoglycan and a free hyaluronic acid chain. Adapted from: (Souza-Fernandes et al. 2006)
2.2.1 Chondroitin sulfate synthesis
Chondroitin/dermatan sulfate GAGs are composed of chains of alternating N-
acetylgalactosamine (GalNAc) and glucuronic acid (GlcA) or iduronic acid (IdoA) residues, and can consist as many as 100 repeating disaccharide units (Silbert and Sugumaran 2002).
Chondroitin sulfates serve as an important structural component of cartilage, providing resistance to compression. The specificity for the synthesis of CS/DS GAGs is provided by the transfer of the first GalNAc residue after completion of the linker tetrasaccharide. The basis for this commitment, and as a result, the emergence of cell-specific GAG chains with a variety of distinct sulfation patterns is not yet fully understood. However, an enzyme that is involved in the transfer of the first GalNAc residue is different from the enzymes that transfer GalNAc in the polymerization process (Ogawa et al. 2010; Sugahara and Kitagawa 2000). Peptide sequence motifs close to GAG bound serine residues could influence the addition of a GalNAc or GlcNAc units to initiate the polymerization of a particular type of GAG chains (Esko and Zhang 1996). CS/DS chain elongation or polymerization is carried
35
out by the action of two distinct transferases involved in the alternating transfer of GalNAc and GlcA (Kitagawa et al. 2001; Silbert and Sugumaran 2002). Six glycosyltransferases are known to be involved in CS synthesis: chondroitin sulfate synthase 1-3 (CSS)1-3,
chondroitin sulfate glucuronyltransferase , and chondroitin N-
acetylgalactosaminyltransferase-1 and-2 (Kitagawa et al. 2001; Sato et al. 2003; Yada et al.
2003a; Yada et al. 2003b), all containing an N-terminal transmembrane domain. (CSS)1-3 exhibit both N-acetylgalactosaminyltransferase-II (GalNAcT-II) and glucuronyltransferase- II (GlcAT-II) dual enzymatic activities (Ogawa et al. 2010). Chondroitin sulfate
glucronyltransferase, on the contrary, displays only GlcAT-1 activity despite having two glycosyltransferase domains (Gotoh et al. 2002). CS GAG chains acquire a variety of modification patterns by the transfer of sulfate groups to hydroxyl groups of both GalNAc and GlcA residues. These modifications are mediated by the action of different STs, giving rise to non-, mono-, di-, or tri- sulfated disaccharides (Kusche-Gullberg and Kjellen 2003).
For example, CS can be sulfated as GalNAc 4-sulfate, as GalNAc 6-sulfate, with both GalNAc 4-sulfate and GalNAc 6-sulfate in the same GAG chains, and occasionally as GalNAc 4,6-disulfate, and also as GlcA 2- or 3- sulfate in combination with GalNAc 4- sulfate or GalNAc 6-sulfate (Silbert and Sugumaran 2002). Epimerization is another modification that occurs to GAG chains as a result of the conversion of GlcA residues at C-5 by chondroitin glucuronate C-5 epimerase (Maccarana et al. 2006) to iduronic acid (IdoA) residue at the polymer level, constituting the formation of dermatan sulfate (DS) (Kjellen and Lindahl 1991; Silbert and Sugumaran 2002; Victor et al. 2009). The hydroxyl groups of GalNAc and IdoA in dermatan sulfate (DS) disaccharides may be substituted by sulfate groups as follows: GalNAc at C-4 and IdoA at C-2 in combination with both non-sulfated GalNAc and IdoA and some sulfated GalNAc at C-6 (Ernst et al. 1995; Saito et al. 1968).
2.2.2 Heparan sulfate synthesis
Heparan sulfate (HS)/heparin contains a carbohydrate backbone constituted by alternating GlcNAc and GlcA or IdoA sugars. As already discussed above, after the completion of the tetrasaccharide linker region synthesis, the polymerization of HS chains is initiated by the addition of a GlcNAc residue by GlcNAc transferase I (GlcNAc-TI) before polymerization proceeds through the alternate transfer of GlcA and GlcNAc sugars, catalyzed by two
36
complexed glycosyl transferases, EXT1 and EXT2 (Hereditary multiple exostoses gene products) (Busse et al. 2007; Carlsson et al. 2008; Sasisekharan and Venkataraman 2000).
Modification of HS chains occurs while the chain is being elongated. The nascent chain is modified concomitantly and sequentially by the action of a series of enzymes (Sasisekharan and Venkataraman 2000). The first modification of the HS chain is initiated by N-
deacetylation and N-sulfation of GlcNAc units by the action of the bifunctional enzyme N- deacetylase/N-sulfotransferase (NDST), requiring PAPS as the sulfate donor (Busse et al.
2007; Carlsson et al. 2008; Salmivirta et al. 1996). Four NDST enzymes (NDST1-4) have been identified (Kusche-Gullberg and Kjellen 2003), with NDST1 and NDST2 being the most widely expressed (Grobe et al. 2002). Subsequent HS chain modifications include epimerization of GlcA to IdoA carried out by HS glucuronyl C-5 epimerase (Li et al. 1997), and 2-O-sulfation of iduronic acid and 6-O and 3-O -sulfation of N-acetylated and N- sulfated GlcN residues (Busse et al. 2007), involving several STs. These modifications are largely confined to N-sulfated regions of the HS chains, indicating the importance of bifunctional NDST enzymes in determining the final structure of HS chains (Carlsson et al.
2008). The magnitude of these modifications varies and results in the generation of heterogeneous HS domains. In comparison with HS GAG chains, heparin is more intensely sulfated and contains a larger proportion of N-sulfated GlcN and IdoA (Kusche-Gullberg et al. 1998). Heparin is defined by the 3-O-sulfate that allows binding of antithrombin, thus the anticoagulant activity.